I bought this mobo in late 2001 as my second PC mobo ever (I was Amiga 4000 PPC user only back in 2001...) and it disapointed me a lot with the overclock. Well, in fact, it was the Celeron 900, witch can't get over 1.02Ghz stable, but that is ancient history now. Now the board serving me as folding + data storage machine. Not long ago I get first crashes. The stability got better ater I exchanged the AGP card to PCI one thanks to better cooling - I suggest looking there:
http://ax2.old-cans.com/s.php?p=wc&id=164&c=8&d=1&v=v2
http://ax2.old-cans.com/s.php?p=wc&id=162&c=8&d=1&v=v2
http://ax2.old-cans.com/s.php?p=wc&id=163&c=8&d=1&v=v2
and there:
http://ax2.old-cans.com/s.php?p=wc&id=161&c=8&d=1&v=v2
to understand what wrong case this is (PSU over the CPU) and witch crazy cooling - CNPS6000Cu with bottom 92mm Zalman fan - I invented to cool the machine down. It sucesfully worked at 120x14 with 1400Mhz Tualatin O/C to 1680MHz and 1.625Vcore
But then I discovered the ughly smell of Teapons, witch suggest that I should replace them at once.
Quite frankly, I was surprised how few caps are out there and how much caps are in Vcore... There is only 7x 100uF 16V caps, 7x 1500uF 6.3V caps and 10x 2200uF 6.3V caps on the Vcore. That is all - the 10uF caps aren't worth replacing or even mentioning, I think
I should admit I did some first ughly cooling attempts on the mobo as well:
Anyway, since Teapos are everywhere:
I decided to end the unreliability and smell by replacing them by Panasonic FM ones.
Abit ST6R
---------
10x 2200uF 6.3V r10 - P12344-ND
7x 1500uF 6.3V r8 - P12343-ND - r10!
7x 120uF 16V r6 - P12922-ND
The Vcore is 2 phase one, but very well build - that could be the answer why Teapos - a known VERY VERY BAD cap's survived 5 years in service with just minor unreliability at the end...
It went well and easy for the Vcore and smaller caps, yet these 1500uF ones are in two cases too close to dimm slot and each other that I could push a r10mm ones (as Pannyes are) there. Luckily, I got few spare hi-quality caps (almost waste use them on that board...) that meet the r8mm requirments - Samxons GC. Samxons produce also r10 and 16mm big GC 1500uF caps I used already with excelent results on the FX5600XT, so...
So I used em:
...creating a sort-of mixed recap
Anyway, I noticed that into the socket, a 3 SMD cap's are omited:
So I added a 5.6uF SMD caps there as well, as I did my "traditional" removing of the LPT and COM ports for the sake of better ventilation
Results.
Smell is gone, stability is back, but I won't exchange the PCI GF4MX440 to any AGP card for the sake of good cooling, of course.
And how's overckock? Good.
The board had before pretty simple and repeatable pattern of O/C with originally 100Mhz FSB CPU's. It goes that way:
120Mhz - stable
122Mhz - semistable
124Mhz - BSOD on win boot
126Mhz - won't post
After replacing caps with no additional changes the board not only posted at 126Mhz FSB (1764MHz for the CPU, BTW...) but it also booted to windows and only after some priming it crashed... That could be attributed to insuficient chipset cooling, as you can check on the links of the older pictures, the chipset cooler is a joke... So, the O/C capabilities of the mobo went up a lot
It looked stable at 122Mhz now with just 1.600Vcore - so the improve on the caps let me O/C higher a bit with even lower Vcore. Cool, is it not?
http://ax2.old-cans.com/s.php?p=wc&id=164&c=8&d=1&v=v2
http://ax2.old-cans.com/s.php?p=wc&id=162&c=8&d=1&v=v2
http://ax2.old-cans.com/s.php?p=wc&id=163&c=8&d=1&v=v2
and there:
http://ax2.old-cans.com/s.php?p=wc&id=161&c=8&d=1&v=v2
to understand what wrong case this is (PSU over the CPU) and witch crazy cooling - CNPS6000Cu with bottom 92mm Zalman fan - I invented to cool the machine down. It sucesfully worked at 120x14 with 1400Mhz Tualatin O/C to 1680MHz and 1.625Vcore

Quite frankly, I was surprised how few caps are out there and how much caps are in Vcore... There is only 7x 100uF 16V caps, 7x 1500uF 6.3V caps and 10x 2200uF 6.3V caps on the Vcore. That is all - the 10uF caps aren't worth replacing or even mentioning, I think

I should admit I did some first ughly cooling attempts on the mobo as well:
Anyway, since Teapos are everywhere:
I decided to end the unreliability and smell by replacing them by Panasonic FM ones.
Abit ST6R
---------
10x 2200uF 6.3V r10 - P12344-ND
7x 1500uF 6.3V r8 - P12343-ND - r10!
7x 120uF 16V r6 - P12922-ND
The Vcore is 2 phase one, but very well build - that could be the answer why Teapos - a known VERY VERY BAD cap's survived 5 years in service with just minor unreliability at the end...
It went well and easy for the Vcore and smaller caps, yet these 1500uF ones are in two cases too close to dimm slot and each other that I could push a r10mm ones (as Pannyes are) there. Luckily, I got few spare hi-quality caps (almost waste use them on that board...) that meet the r8mm requirments - Samxons GC. Samxons produce also r10 and 16mm big GC 1500uF caps I used already with excelent results on the FX5600XT, so...
So I used em:
...creating a sort-of mixed recap

Anyway, I noticed that into the socket, a 3 SMD cap's are omited:
So I added a 5.6uF SMD caps there as well, as I did my "traditional" removing of the LPT and COM ports for the sake of better ventilation

Results.
Smell is gone, stability is back, but I won't exchange the PCI GF4MX440 to any AGP card for the sake of good cooling, of course.
And how's overckock? Good.
The board had before pretty simple and repeatable pattern of O/C with originally 100Mhz FSB CPU's. It goes that way:
120Mhz - stable
122Mhz - semistable
124Mhz - BSOD on win boot
126Mhz - won't post
After replacing caps with no additional changes the board not only posted at 126Mhz FSB (1764MHz for the CPU, BTW...) but it also booted to windows and only after some priming it crashed... That could be attributed to insuficient chipset cooling, as you can check on the links of the older pictures, the chipset cooler is a joke... So, the O/C capabilities of the mobo went up a lot

It looked stable at 122Mhz now with just 1.600Vcore - so the improve on the caps let me O/C higher a bit with even lower Vcore. Cool, is it not?

Comment